Journal of Economic Perspectives—Volume 20, Number 3—Summer 2006 —Pages 71–96
The Growth in the Social Security
Disability Rolls: A Fiscal Crisis
Unfolding
David H. Autor and Mark G. Duggan
M
ore than 80 percent of all nonelderly adults in the United States are
insured against the risk of being unable to work because of a physical or
mental disability by the Social Security Disability Insurance (DI) program. To be insured for DI benefits, a person must have worked in a job covered
by Social Security in at least five of the ten most recent years. To be awarded
benefits, an individual must have a medically determinable physical or mental
impairment that is expected to result in death or to last for at least a year and that
prevents the person from engaging in a “substantial gainful activity.”
During the past two decades, the fraction of individuals receiving Disability
Insurance benefits has grown substantially, as shown in Figure 1. In 1985,
2.2 percent of individuals between the ages of 25 and 64 were receiving DI
benefits, but by 2005 this fraction had risen to 4.1 percent. If recent entry and
exit rates continue in the years ahead, then more than 6 percent of the
nonelderly adult population will soon be receiving DI benefits.
The rapid expansion of the beneficiary population has three main causes.
First, a set of congressional reforms in 1984 to Disability Insurance screening led to
rapid growth in the share of recipients suffering from back pain and mental illness.
Because these disorders have comparatively low mortality, the average duration of
disability spells—and hence the size of the recipient population— has increased.
Second, a rise in the after-tax DI income replacement rate—that is, the ratio of
disability income to former labor earnings—strengthened the incentives for work-
y David Autor is Associate Professor of Economics, Massachusetts Institute of Technology,
Cambridge, Massachusetts. Mark Duggan is Associate Professor of Economics, University of
Maryland, College Park, Maryland. Both authors are Faculty Research Fellows, National
Bureau of Economic Research, Cambridge, Massachusetts. Their e-mail addresses are
具[email protected]典 and 具[email protected]典.
72
Journal of Economic Perspectives
Figure 1
Disability Insurance Recipiency and Award Rates per Adult Ages 25– 64,
1957–2005
(excludes dependents)
Percent receiving benefits
4.0%
0.7%
0.6%
3.5%
0.5%
3.0%
2.5%
0.4%
2.0%
0.3%
1.5%
Award rate per population
4.5%
0.2%
1.0%
0.5%
0.0%
19
57
19
60
19
63
19
66
19
69
19
72
19
75
19
78
19
81
19
84
19
87
19
90
19
93
19
96
19
99
20
02
20
05
0.0%
0.1%
Source: Social Security Bulletin: Annual Statistical Supplement (various years) and U.S. Census Bureau.
ers to seek benefits. Third, a rapid increase in female labor force participation
expanded the pool of insured workers. The aging of the baby boom generation has
contributed little to the rise of receipt of disability benefits, while improvements in
population health have likely reduced the incidence of disabling medical disorders.
The growth in Disability Insurance receipt has substantial fiscal ramifications
for the health of the Social Security system and the federal budget as a whole. Since
1985, the fraction of Social Security spending accounted for by the DI program rose
from 10 to 17 percent and the payroll tax devoted to DI increased from 1.0 percent
to 1.8 percent. Real expenditures on the DI program have increased more rapidly
than enrollment. In 2005, the Social Security Administration disbursed
$85.4 billion to 6.5 million DI recipients and their families. Most of these
benefits went to disabled persons, with almost $6 billion paid to the 1.6 million
eligible children of DI recipients and an additional $0.5 billion paid to the 0.16
million eligible spouses of DI recipients.
More than 832,000 persons were newly awarded Disability Insurance benefits
in 2005. To gauge the fiscal commitment these awards entail, consider an average
case: a 50-year-old DI awardee who has a constant 3.1 percent annual mortality rate,
receives an average monthly benefits equal to $1100, and only receives DI benefits
until age 66 (when Social Security retirement benefits kick in). If future benefits are
David H. Autor and Mark G. Duggan
73
discounted at a real 2 percent annual rate, the present value of this average DI
award is $150,000. This implies that the 832,000 DI awards made during the 2005
calendar year had a present value of approximately $125 billion. DI recipients are
also eligible for health insurance through the Medicare program. The cost of
providing this care was $49 billion during the 2005 fiscal year, which amounted to
$7,700 per recipient. Taking into account the value of these benefits until age 65
(after which the time claimant would become eligible for Medicare even without a
DI award), the present value of federal spending for the average DI awardee
increases to more than $245,000.1 Using current population and cost figures, the
steady-state cost of these benefits is $204 billion dollars annually, equal to
8.1 percent of the $2.5 trillion federal budget in 2005 and five times the cost of
the federal Department of Homeland Security. While the rate of DI receipt is lower
in the United States than in other developed Western economies, the American
program has grown dramatically over the last 20 years in size and expense.2 This
growth poses significant risks to the finances of the DI program and the broader
Social Security system, and raises troubling questions as to whether the program is
being misused by claimants.
This article begins by providing an overview of the Disability Insurance program, describing who qualifies for the program, how an individual applies for
benefits and how the level of benefits is determined.3 Next, we summarize the
factors that are responsible for the growth in the DI rolls and discuss how the
characteristics of DI recipients have changed as a result. We then explore the extent
of moral hazard in the DI program and the effectiveness of the screening process
in distinguishing meritorious from nonmeritorious claims. The subsequent section
identifies the challenges that the DI program creates for Social Security finances
and Social Security reform. In the final section, we discuss potential reforms to the
DI program.
An overarching issue raised by our discussion is that five decades of congressional reforms and extensions of Disability Insurance have left the program with an
ill-defined mission. Although the program nominally exists to insure workers’ labor
1
This calculation assumes that the real value of Disability Insurance benefits remains constant for
16 years and 4 months (a 50-year-old born in 1956 would have a full retirement age of 66 and 4 months),
which seems reasonable given that benefits are adjusted each year by the growth in the Consumer Price
Index. Medicare spending per recipient is assumed to grow at 2.3 percent per year, the average rate from
2003 to 2005, though only the first 15 years are considered because the person would otherwise have
been eligible on his or her 65th birthday.
2
According to OECD (2003), the fraction of nonelderly adults receiving disability benefits is significantly lower in the United States than in most other industrialized countries. The average rate of
enrollment among adults ages 20 to 64 in other OECD-17 countries is 6.5 percent, versus 4.5 percent in
the United States. Additionally the ratio of average disability benefits to per-capita wages in 1999 was
lower in the United States than in any other OECD-17 country.
3
The federal government provides disability insurance through other programs, too. For example, the
means-tested Supplemental Security Income (SSI) program paid disability benefits to 4.1 million
non-elderly adults, 1.0 million children, and 800,000 elderly adults in 2004. The other major programs
are administered by the Department of Veterans Affairs, which currently pays benefits to more than
3.0 million veterans of military service.
74
Journal of Economic Perspectives
income against disabling medical events, the DI screening procedure put in place
by Congress hinges to a significant extent on an applicant’s employability, not just
personal health, causing the program to function much like a long-term unemployment insurance program for the unemployable. Ultimately, we believe that
Congress must clarify whether the primary social objective of the DI program is to
provide disability insurance or to provide employability insurance—and if the
former, to revise the selection criteria accordingly. Short of this objective, we
suggest below that the two most promising near-term opportunities for reform
entail revamping the disability appeals process—in which the Social Security Administration currently loses nearly three-quarters of all appeals—and reducing the
attractiveness of DI benefits for work-capable disabled individuals by providing
additional access to public health insurance.
The Social Security Disability Insurance Program
Social Security began paying monthly retirement benefits to workers over the
age of 65 in 1940, but disability benefits were not added until 1956. At their
introduction, Congress conceived of disability benefits as providing early retirement insurance for the “totally and permanently disabled,” to use the phrasing of
the 1950 grant-in-aid program for state public assistance on which the Disability
Insurance program was based. As such, benefits were limited to the near-elderly,
ages 50 to 64, who were unable to “engage in any substantial gainful activity by
reason of any medically determinable physical or mental impairment which can be
expected to result in death or to be of long-continued and indefinite duration.”
Disabled adult children of deceased, retired or disabled workers could also qualify
for DI if they had become disabled before reaching the age of 18.
During the next 30 years, the narrow scope of the Disability Insurance program
was broadened substantially. In 1960, DI benefits were extended to workers under
the age of 50. In 1965, the definition of disability was expanded to include
impairments expected to last at least a year, thus relaxing the requirement that
disabilities must be of “long-continued and indefinite duration.” In 1973, DI
beneficiaries became eligible for Medicare (the health insurance program for the
elderly) after 24 months of disability. In 1984, Congress enacted reforms (detailed
below) that substantially liberalized the disability screening progress, thus making
disability benefits considerably more accessible to workers with non-life-threatening
disorders such as mental illness and back pain. Cumulatively, these programmatic
changes expanded the original narrow mandate of the DI program to encompass
a broader population with a less precisely defined entitlement to benefits.
Adding to the complexity of an expanding program mission, five decades of
advances in medical treatments and rehabilitative technologies, combined with a
secular trend away from physically exertive work, have arguably blurred any sharp
divide that may have once existed between those who are “totally and permanently
disabled” and those who are disabled but retain some work capacity. While one
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
75
might have expected these medical and labor market changes to reduce the
incidence of disabling medical conditions and hence lower the relative size of the
DI program, this has not occurred.
The Application Process
To apply for disability benefits, an individual must submit an application to a
Social Security Administration field office, of which there are currently approximately 1,400 nationwide. Employees at the field office first check nonmedical
criteria. Only adults below the full retirement age—which was 65 for individuals
born in 1937 or earlier, and then phased up over time to 67 for those born in 1960
or later—are eligible for Disability Insurance. The applicant must have worked in
at least five of the ten most recent years and cannot currently be engaging in a
substantial gainful activity (no more than $860 in monthly earnings in 2006). In an
average year during the 1990s, approximately 10 percent of applications were
denied for nonmedical reasons.
If the applicant meets the nonmedical criteria, the application is forwarded to
one of the Disability Determination Services offices (Social Security Administration,
2004). These offices are administered by state governments, but act as agents of the
Social Security Administration (House Ways and Means Committee, 1994). At this
office, disability examiners and medical staff scrutinize medical evidence from one
or more of the applicant’s health care providers regarding the applicant’s ability to
perform work-related activities. If this medical information is insufficient, the office
will commission and pay for an outside consultative medical examination of the
applicant. If the Disability Determination Services office finds that the applicant has
an impairment that meets or exceeds one of those on the official Listing of
Impairments, then a disability award is made. If not, the next step is to consider the
applicant’s ability to work, taking into account health status, age, education, work
experience and the transferability of the applicant’s skills. If the office reaches a
conclusion that the applicant cannot work, a disability award is made. In a typical
year during the 1990s, approximately 37 percent of applicants were awarded
benefits at this first stage.
If the original Disability Insurance application is denied, an applicant can
pursue three levels of appeals, in this order: 1) request a reconsideration by a
different team at the state office of Disability Determination Services; 2) request a
hearing with an Administrative Law Judge; 3) appeal to Social Security’s Appeals
Council, to the U.S. District Court and finally to the U.S. Circuit Court of Appeals.
Of those initially denied benefits in recent years at the first stage of the application
process, approximately 58 percent filed at least one appeal. More than half of those
who appealed were eventually awarded disability benefits.
The Generosity and Duration of Disability Insurance Benefits
The level of disability benefits received is determined by a formula that starts
by calculating average indexed monthly earnings (AIME), using the earnings data
that is collected for calculating Social Security benefits. Past wages are indexed up
76
Journal of Economic Perspectives
to the present using an “inflator” equal to average wage growth in the economy.4
Years with the lowest earnings are excluded from the average, with five years after
the age of 21 excluded for all awardees ages 43 and up and fewer years excluded for
younger recipients. For very low-wage workers, disability benefits are equal to
90 percent of average indexed monthly earnings. As average indexed monthly
earnings rise, the proportion of income replaced declines, so that low-wage workers
replace a larger fraction of their earnings than do high-wage workers.5 The average
monthly benefit for newly disabled workers in 2005 was $970, and close to $1,100
when accounting for dependent benefits. As noted above, two years after the onset
of their disability, recipients of Disability Insurance benefits also receive health
insurance coverage through the federal Medicare program.
Once a person is enrolled in Disability Insurance, there are three main
pathways out of the program. First, the DI recipient may reach the full retirement
age, at which point the claimant shifts to the retired worker portion of Social
Security. In 2004, more than 44 percent of those who left DI did so for this reason.
Second, the claimant may die, which accounted for 42 percent of all program exits
in 2004. Third, the claimant may no longer meet the nonmedical or medical
standards for receiving disability benefits. For example, an individual may return to
work and earn more than the permitted minimum amount, or the Social Security
Administration may conduct a Continuing Disability Review and determine that the
individual is no longer disabled. During the 2004 calendar year, eligibility-based
exits accounted for 12 percent of all exits from the DI program.
In 1985, the Disability Insurance exit rate stood at 12.1 percent—that is, one
in eight claimants left the rolls. It has trended downward steadily since that time,
reaching a low of 7.2 percent in 2004, an exit rate of just one in fourteen claimants.
4
Nominal earnings in year t for a person applying in year T are multiplied by the ratio of average wages
in year T ⫺ 2 to average wages in year t, if t ⫽ T ⫺ 2 or earlier. Indexed earnings equal nominal
earnings in year T ⫺ 2. Indexed earnings are set equal to nominal earnings for years T ⫺ 1, and T.
5
Specifically, an individual who becomes disabled at age A years qualifies for DI average indexed
monthly earnings (AIME) defined as:
AIME ⫽
冉
冊冘
1
ⴱ
A ⫺ 25
A
a⫽21
冉
冊
Y៮ A⫺2 ⴱ
Y a ⴱMax 1, ៮
Ia .
Ya
In this equation, Y a equals the person’s average monthly earnings that were subject to Social Security
taxes at age a and Y៮ a equals average economy-wide earnings in that same year. Years with low earnings
are excluded from this calculation, as explained in the text. The indicator variable I a is equal to one if
indexed earnings at age a are included in this calculation and zero otherwise. The monthly DI benefit
is computed from the AIME using the following formula:
PIA ⫽
再
0.9 * AIME
0.9 * b 1 ⫹ 0.32 * 共AIME ⫺ b 1兲
0.9 * b l ⫹ 0.32 * 共b2 ⫺ b1 兲 ⫹ 0.15* 共AIME ⫺ b2 兲
if
if
if
AIME 僆 关0, b 1兴
AIME 僆 关b 1, b 2兴 ,
AIME ⬎ b2
where the bend points b 1 and b 2 are scaled each year by average nominal wage growth in the economy.
David H. Autor and Mark G. Duggan
77
120
180
Total terminations per 1,000 recipients
160
100
140
80
120
100
60
80
40
60
40
20
20
0
0
1963 1967 1971 1975 1979 1983 1987 1991 1995 1999 2003
Terminations per 1,000 recipients due to:
Medical disqualifications
Deaths
Retirements
Figure 2
Disability Insurance Termination Rates per 1,000 Beneficiaries by Reason,
1963–2004
Source: Social Security Bulletin: Annual Statistical Supplement (various sources).
As shown in Figure 2, this decline has been driven both by a drop in the fraction
of DI recipients reaching the full retirement age (from 6.7 to 3.3 percent) and in
the fraction dying during the year (from 4.9 to 3.1 percent).
Why Are the Disability Rolls Growing?
By far the most important contributor to the recent growth of Disability
Insurance is the liberalization of the DI screening process in 1984, which increased
the number of disability awards and shifted the composition of recipients towards
claimants with lower mortality disorders. Two other substantial contributors are
rising financial incentives to apply for a disability award and changes in labor force
participation that increase the share of citizens insured by, and therefore eligible to
receive, DI benefits. By contrast, changes in health and age structure do not appear
to be quantitatively important contributors. We discuss these factors in turn.
Changes in Disability Screening
The Disability Insurance program first paid out benefits in 1957, with 150,000
disabled workers enrolled in December of that year. By the end of 1977, 2.8 million
nonelderly adults (2.8 percent of the nonelderly adult population) received dis-
78
Journal of Economic Perspectives
abled worker benefits from the DI program and an additional 2.0 million spouses
and children received dependent benefits. The unexpectedly rapid increase in DI
enrollment during the 1970s led to a funding crisis, with expenditures exceeding
revenues by 25 percent in 1977 and the (inflation-adjusted) balance in the DI trust
fund declining by more than 65 percent from 1974 to 1977. As a result, the
state-administered Disability Determination Services offices were instructed to
tighten the medical eligibility criteria for the program. Initial denial rates increased
by 15.5 percentage points (30 percent) during the next three years (Gruber and
Kubik, 1997). The medical eligibility criteria were tightened further by 1980 federal
legislation that substantially increased the number of continuing disability reviews
and made it more difficult for applicants to qualify for benefits. Between 1980 and
1983, the Social Security Administration determined that more than 380,000
beneficiaries of DI— 40 percent of those whose cases were reviewed—no longer
met medical standards and then it terminated their benefits (Rupp and Scott,
1998).
These actions yielded a significant decline in applications, awards and enrollment from 1980 to 1983—and also generated a public backlash. Congress responded with legislation in 1984 that made changes in two broad areas, profoundly
altering the medical eligibility criteria for the Disability Insurance program and
changing the evidentiary rules for continuing disability reviews. Prior to 1984, the
disability determination process had focused on whether an applicant’s objectively
verifiable diagnostic criteria met or exceeded a listed impairment. The 1984
legislation enlarged this focus to consider an applicant’s “ability to function in a
work-like setting.” Under this law, the Social Security Administration was to place
significant weight on applicants’ reported pain and discomfort, to relax its strict
screening of mental illness and to consider multiple nonsevere ailments (“impairments”) as constituting a disability during the initial determination decision, even
if none of these impairments was by itself disabling. Though these reforms did not
alter the statutory definition of disability, they shifted the focus of screening from
medical to functional criteria, which had a major effect on decision making. In the
early years of the DI program, 93 percent of initial awards were based strictly on
medical factors. In the year prior to the 1984 reforms, this number was 82 percent.
Seventeen years later, the share of initial awards based strictly on medical factors
had plunged to 58 percent (Social Security Advisory Board (SSAB), 2003).
The second major component of the 1984 legal reform made evidentiary rules
for determining disability status more favorable to claimants. Historically, the Social
Security Administration had placed greatest weight on its own consultative medical
examination. The 1984 regulations required that evidence provided by the applicant’s own health care provider have “controlling” weight, provided this evidence
was not at odds with other medical evidence. These changes gave far more scope for
appeal, with a resulting further decline in the share of awards due to strictly medical
factors. While 58 percent of initial awards in 2000 were based on strictly medical
factors, by the end of the appeals process, the share of all awards due to purely
medical factors was only 40 percent (SSAB, 2003).
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
79
Table 1
Disability Insurance Awards by Diagnosis Group, 1983 and 2003
Number of DI
awards
Diagnosis group
Heart disease
Cancer
Mental disorders
Musculoskeletal disorders (e.g. back pain)
Nervous system
Respiratory system
Injuries
Endocrine system (e.g., diabetes)
All other
Total
Awards per
1,000 insured
Share of awards
1983
2003
1983
2003
1983
2003
68,352
52,379
50,633
41,782
26,203
17,978
15,646
14,904
23,613
311,490
85,896
70,942
191,679
199,014
64,369
32,007
28,612
23,407
59,780
755,706
0.65
0.49
0.48
0.39
0.25
0.17
0.15
0.14
0.22
2.94
0.60
0.49
1.33
1.38
0.45
0.22
0.20
0.16
0.42
5.25
21.9%
16.8%
16.3%
13.4%
8.4%
5.8%
5.0%
4.8%
7.6%
100.0%
11.4%
9.4%
25.4%
26.3%
8.5%
4.2%
3.8%
3.1%
7.9%
100.0%
Source: Data on awards by diagnosis obtained from the 2003 “Annual Statistical Report on the Social
Security Disability Insurance Program” as shown here: ⬍http://www.ssa.gov/policy/docs/statcomps/
di_asr/2003/index.html⬎. Data on the number of people insured for DI obtained from the SSA’s Office
of the Actuary.
Note: Award rate defined as awards per 1000 individuals insured for DI in each year.
The reform of Disability Insurance in the 1980s ushered in two decades of
program expansion. In the 21 years following the 1984 legislation, the DI rolls grew
by 148 percent, from 2.6 million to 6.5 million beneficiaries, and the percentage of
the nonelderly adult population receiving disability benefits from the program
grew from 2.2 to 4.1 percent. For comparison, the number of workers insured by
Disability Insurance grew by 38 percent and the nonelderly U.S. adult population
ages 25 to 64 grew by 33 percent.
Accompanying this numerical growth was a major shift in the composition of
the Disability Insurance beneficiary population. Table 1 shows that just four categories of diagnoses (“impairments”) make up approximately 70 percent of all
disability awards: heart disease (more generally, circulatory disorders); cancers;
mental impairments; and musculoskeletal disorders (typically back pain or arthritis). Since 1983, the relative importance of these categories has changed dramatically. Reflecting the increased weight given to pain and workplace function, the
number of new DI awardees with a primary diagnosis of musculoskeletal disease or
mental disorder—the two disorders with the lowest mortality among the 14 major
diagnostic categories (Hennessey and Dykacz, 1993)—increased by 323 percent
between 1983 and 2003 and accounted for more than half of all disability awards in
the latter year. The corresponding increases in cancers and heart disease awards
together were 30 percent, and thus lower than the increase in the insured population during that period.
Because mental and musculoskeletal disorders have an early onset and low
age-specific mortality, Disability Insurance beneficiaries with these diagnoses
80
Journal of Economic Perspectives
experience relatively long durations on the program. Thus, in 1983, 4.9 percent of
people receiving disability benefits in that year died; by 2004, only 3.1 percent of
those receiving disability benefits in that year died. Reflecting the younger age of
entry into DI, the fraction exiting the program for retirement declined from
7.0 percent in 1983 to just 3.3 percent by 2004, as shown in Figure 2. By implication,
the average duration of receipt of disability benefits among new beneficiaries has
lengthened substantially as the share leaving the program due to death or retirement has diminished.
The Rising Value of Disability Insurance Benefits
The value of Disability Insurance benefits relative to potential labor market
earnings—the “replacement rate”— has risen substantially since the late 1970s
(Autor and Duggan, 2003). This rise is not the result of a direct legislative decision,
but rather has occurred because of an interaction between the disability benefits
formula and the growth of earnings inequality in the U.S. economy (Katz and
Autor, 1999). Although DI benefits awarded are nominally only a function of a
worker’s prior earnings, award amounts are calculated using a wage index equal to
mean wage growth economy-wide. Consequently, an individual’s benefit also
depends implicitly upon the individual’s earnings growth relative to the growth of
earnings for all workers during that worker’s years of employment.
Figure 3 illustrates how this indexation scheme interacts with earnings inequality to raise the replacement rate of low-earnings workers. Line segment A-B-C
depicts the benefits schedule of a worker awarded Disability Insurance benefits in
1980 whose wage growth prior to receiving DI exactly paced mean earnings in the
economy. The worker’s calculated average indexed monthly earnings amount
(AIME) is identical to her 1980 wage. Because the benefits formula replaces
between 15 and 90 percent of the marginal dollar (depending on the claimant’s
AIME), her monthly payment Primary Insurance Amount (PIA) falls somewhat
below her 1980 wage.
Next, consider a worker, represented by line segment A-B-D-E, who is awarded
Disability Insurance benefits in 2000. This worker’s nominal wage history in 2000 is
identical to that of the beneficiary in 1980 but, in contrast to the 1980 beneficiary,
his wage growth during his career lagged contemporaneous annual average wage
growth economy-wide. This worker will receive a higher real Primary Insurance
Amount than the worker entering Disability Insurance in 1980. Why? The indexation
of the earnings “brackets”—that is, the ranges over which income is replaced at the 90,
32 or 15 percent rates—moves these brackets upward, causing a larger share of the
worker’s income to be replaced at the 90 or 32 percent rates than would have been
the case in 1980. We label this as the “bracket effect” in Figure 3. Indexation also
raises this worker’s DI benefit through a second channel. Because the more recent
worker’s entire earnings history is inflated by historical mean wage growth, his
average indexed monthly earnings amount will actually exceed current earnings
(recall that his wage growth has lagged the economy-wide average). We label this as
the “earnings history effect” in Figure 3. Jointly, these two forces—indexation of the
David H. Autor and Mark G. Duggan
81
Primary Insurance Amount (PIA$)
Figure 3
Illustration of the Impact of Earnings Inequality and Indexation on Disability
Insurance Benefits in 1980 and 2000
E
PIA 2000
Earnings history effect
C
D
Bracket effect
PIA 1980
B
Marginal
replacement
rate ⫽ 90%
A
$0
Bend 1980
Marginal
replacement
rate ⫽ 32%
Bend 2000
AIME 1980
AIME 2000
Average Indexed Monthly Earnings (AIME$)
Note:
PIA ⫽
再
0.9 * AIME
0.9 * b 1 ⫹ 0.32 * 共AIME ⫺ b 1兲
0.9 * b l ⫹ 0.32 * 共b2 ⫺ b1 兲 ⫹ 0.15* 共AIME ⫺ b2 兲
if
if
if
AIME 僆 关0, b 1兴
AIME 僆 关b 1, b 2兴 ,
AIME ⬎ b2
where the bend points b 1 and b 2 are scaled each year by average nominal wage growth in the economy.
earnings brackets and indexation of past earnings— have substantially raised the
income replacement rate of low-earnings DI beneficiaries since 1979, when earnings inequality began growing rapidly.
Augmenting the increase in the replacement rate for Disability Insurance
benefits is a substantial rise in the real value of the Medicare entitlement received
by disabled persons. Average Medicare expenditures per recipient of DI more than
doubled from $3,259 in 1979 to $7,700 in 2005 (both in 2005 dollars).6 Simultaneously, private sector health insurance coverage rates fell steeply, particularly for
low-earnings workers (Farber and Levy, 2000), thus further increasing the relative
value of the Medicare benefit.
Table 2 shows the effects of these factors on Disability Insurance replacement
rates for currently employed male workers ages 30 to 61 at various percentiles of the
age-specific earnings distribution in 1984 and 2002. The first pair of columns shows
6
Bhattacharya and Schoenbaum (2002) estimated the present value of the Medicare benefit to new
Disability Insurance recipients at $40,000 for males and $50,000 for females, with a value as high as
$75,000 for recipients with a diagnosis of mental illness.
82
Journal of Economic Perspectives
Table 2
Estimated Changes in Disability Insurance Earnings Replacement
Rates for Males Ages 30 to 61 by Earnings Percentile, 1984 and
2002
Replacement rates adding
fringe benefits, Medicare
Wage replacement rates
1984
2002
10th
50th
90th
48.4%
36.2%
24.1%
59.4%
41.9%
26.1%
10th
50th
90th
51.1%
33.5%
19.4%
55.1%
43.3%
24.8%
10th
50th
90th
55.2%
34.7%
19.0%
64.0%
45.9%
23.7%
1984
2002
60.6%
35.4%
22.5%
85.7%
44.4%
24.7%
62.7%
32.7%
18.4%
76.9%
44.4%
23.3%
67.8%
34.1%
18.2%
86.0%
46.4%
22.4%
Males 30–39
Males 40–49
Males 50–61
Source: Authors’ calculations from the March Annual Demographic Supplement of
the CPS 1964 –2002.
the earnings replacement rate for different groups of retirees. The second pair of
columns shows replacement rates with the value of Medicare benefits taken into
account, while netting out average fringe benefits earned on the job. For example,
in 1984, currently employed male workers ages 50 to 61 at the tenth percentile of
the earnings distribution could potentially have replaced 55 percent of their
current earnings with DI cash transfers. By 2002, this share had risen to 64 percent.
Accounting for the rising value of in-kind Medicare benefits and netting out
average fringe benefits earned on the job in the final column, the overall disabilityrelated replacement rate for a tenth percentile male age 50 to 61 rose still further
from 68 to 86 percent.
Nor was this rise limited to older workers. For males at or below the median
earnings level, the rise in the replacement rate inclusive of Medicare ranged from
9 to 25 percentage points across all age brackets. The rise in the replacement rate
was far less pronounced at higher positions in the earnings distribution, as one
would expect from the way in which benefits are calculated. For workers at the
90th percentile of earnings, potential replacement rates rose a comparatively modest 1 to 2 percentage points.
The rising Disability Insurance replacement rate probably spurred additional
DI applications and awards. While it is difficult to distinguish the effects of changes
in program rules versus changes in generosity of program benefits, the replacement
rate increase was particularly pronounced for low-earnings workers, the group who
saw their real market wages fall sharply during the 1980s. This pattern suggests that
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
83
Table 3
Rate of Disability Insurance Receipt by Gender, Race, Age and Education:
1984 and 2004
Males
Females
1984
2004
1984
2004
Actual DI enrollment
25–64, All
25–39, All
40–54, All
55–64, All
3.0%
0.9%
2.8%
9.4%
4.3%
1.3%
4.3%
10.9%
1.4%
0.4%
1.2%
4.2%
3.5%
1.1%
3.6%
8.3%
SIPP DI enrollment
25–64, All
25–64, White
25–64, Black
2.8%
2.7%
4.6%
3.5%
3.4%
5.5%
1.8%
1.6%
3.5%
3.5%
3.2%
5.4%
25–39, HS dropout
40–54, HS dropout
55–64, HS dropout
2.3%
5.4%
14.8%
2.5%
7.8%
19.7%
1.0%
3.8%
9.1%
2.7%
6.5%
12.7%
25–39, HS graduate
40–54, HS graduate
55–64, HS graduate
0.9%
2.3%
6.7%
2.0%
3.4%
9.6%
0.5%
1.3%
3.6%
1.9%
3.8%
8.1%
25–39, College degree
40–54, College degree
55–64, College degree
0.1%
1.2%
2.7%
0.4%
1.5%
3.7%
0.2%
0.6%
1.8%
0.5%
1.9%
5.0%
Source: Authors’ tabulations from SSA administrative data (first four rows) and the first wave of the 1984
and 2004 versions of the Survey of Income and Program Participation (SIPP) (all other rows).
Note: SIPP samples include adults ages 25 to 64, with sample sizes of 25,591 in 1984 and 57,412 in 2004.
High-school-dropout category includes those who have not completed a 12th grade education or passed
the GED. College-degree category includes those with two or more completed years of schooling in the
1984 survey and those with an associate’s degree or greater in the 2004 survey. The HS-graduate category
includes all other individuals. Age denotes respondent’s age in the first month of the SIPP wave.
DI should have become more attractive to the less-skilled and less-educated over
this period. We examined this implication using data from the Survey of Income
and Program Participation (SIPP), and some of the results are shown in Table 3.
The increase in DI enrollment during the last two decades was largest for those
without a high school degree, with especially large increases for those between the
ages of 40 and 64. In 2004, high school dropout males ages 40 through 64 were five
times as likely as males with a college degree and twice as likely as males with only
a high school degree to receive DI benefits.
The Roles of Aging and Labor Supply
The aging of the U.S. population—in particular, the passing of the baby boom
generation into middle age— has so far made only a modest contribution to the
growth of Disability Insurance. Calculations from Duggan and Imberman (forth-
84
Journal of Economic Perspectives
coming in 2006) that have been updated for this article reveal that, holding
age-specific rates of receipt of disability benefits at their 1984 base, the aging of the
population between 1984 and 2004 explains only 6 percent of the increase in the
fraction of nonelderly adults receiving Disability Insurance.
Why is this contribution so small? First, the baby boom cohorts have yet to
reach their peak disability years. Disability rises rapidly starting around age 50. In
1984, a male age 60 to 64 was 5.7 times more likely to receive Disability Insurance
than a male age 40 to 44 and 1.9 times more likely than a male age 50 to 54 (based
on data from the Social Security Administration Annual Statistical Supplement, 1985,
2005). However, between 1980 and 2000, the share of the population that was
near-elderly, ages 50 to 64, was nearly constant at 15 percent of the population.
A second reason that aging does not appear more important is that its
contribution is numerically overwhelmed by the growth of Disability Insurance
receipt within given age groups. For example, if one divides males between 25 and
64 into five-year age groups, DI receipt increased within each group by an average
of 41 percent. This increase was especially sharp for males ages 40 to 49, for whom
the rate of receipt rose by 65 percent. Ironically, the only male age group for which
disability receipt did not increase by more than 20 percent during this period is
those aged 60 to 64.
The growth of disability receipt between 1984 and 2004 was far more pronounced among women than men. Whereas male disability receipt grew by 41 percent, female receipt grew by 151 percent, thus raising the ratio of female to male
recipients from five females for every ten males in 1984 to eight females for every
ten males in 2004. A significant—though far from complete— explanation for
greater grown of DI receipt among women is the secular rise in female labor supply
in recent decades, which has been particularly rapid since 1970 (Blau, Ferber and
Winkler, 2002). Between 1984 and 2004, the share of women ages 25 to 64 eligible
to receive Disability Insurance benefits—that is, having worked in a job covered by
Social Security in at least five of the ten most recent years— grew by 25 percent,
from 61 to 76 percentage points. Holding rates of receipt of DI constant among
workers at its 1984 level, the increased number of women in the paid labor force
can explain only about one-sixth of the 151 percent increase in the rate of DI
receipt among women. By contrast, the fraction of nonelderly men covered by DI
has remained virtually unchanged in this period (at roughly 90 percent) and so has
not been a major contributor to the increase in their receipt of disability benefits.
The Role of Health
A final potentially important demographic factor affecting the receipt of
Disability Insurance is health. One straightforward and commonly used measure of
the health status of a group is mortality. Between 1981 and 2001, annual mortality
rates for men and women between the ages of 50 and 64 fell by 29 and 17 percent
respectively, according to data from the Social Security Administration available
online at 具http://www.demog.berkeley.edu/⬃bmd/states.html典. Annual mortality
rates for men in their 40s declined by approximately 18 percent during this period
David H. Autor and Mark G. Duggan
85
and for women in this same age group by 24 percent. Perhaps other adverse health
trends, like obesity, may have diminished the ability of working-age adults to
participate in the labor market. Duggan and Imberman (forthcoming in 2006)
explore this issue using measures of self-reported health status from the National
Health Interview Survey, focusing on the prevalence of activity or work-limiting
conditions. Their findings for adults between the ages of 50 and 64 —a group that
accounted for 62 percent of all DI recipients in 2004 —reveal a substantial improvement in their average health since 1984. Reinforcing these conclusions, Manton
and Gu (2001) find that the share of the population age 65 and over suffering from
a chronic disability fell from 26.2 in 1982 to 19.7 percent in 1999 —a decline of 33
percent—with the largest drop between 1994 and 1999 (see also U.S. Census
Bureau, 2005, chapter 3).
In summary, we conclude that changes to the Disability Insurance program
itself—the liberalization of screening and the rise in the replacement rate—and
their interactions with the labor market are the central factors explaining the
outsized growth of Disability Insurance receipt since 1984. Aging and health are
comparatively minor factors.
What Share of Disability Recipients are Undeserving?
The Work Capability of Disability Insurance Recipients
In the last two decades, the average health of the U.S. population has been
improving while the share of working-age adults receiving disability benefits has
been increasing. Judged by historical standards, are all of the new recipients
“deserving” of a lifetime of income and medical care paid for by the Social
Security Administration and the Centers for Medicare and Medicaid Services?
Put more bluntly, are a substantial share of Disability Insurance recipients
cheating? This cheating could take different forms. Work-capable individuals,
including those with nondisabling impairments, might exit the labor force to
claim benefits. Alternatively, individuals who are voluntarily or involuntarily out
of the labor force but not impaired might file a disability claim to gain income
support and health insurance.7
The extent of such cheating is difficult to evaluate for both practical and policy
reasons. At a practical level, there are no systematic, objective data on the work
capacity of current Disability Insurance beneficiaries and so there is no reliable
7
In evaluating the first type of moral hazard wherein workers leave the labor force to claim benefits, one
should ideally distinguish between distortionary and nondistortionary components of moral hazard.
Disability Insurance provides a large permanent income boost in the case of an adverse medical shock,
and this may reduce labor supply purely through an income effect—a nondistortionary effect. Alternatively, the program may cause workers to reduce labor supply to appear more disabled and therefore
qualify for DI benefits. This is a distortionary (substitution) effect. We know of no existing research on
disability that attempts to distinguish these effects (except for work in progress by the authors, Autor and
Duggan, 2006).
86
Journal of Economic Perspectives
means to estimate what share could potentially work. Nor can the question of
cheating be resolved by an appeal to medical standards. While certain medical
conditions are clearly disabling, “disability” is not a medical condition. Disability
is a dividing line (or zone) chosen by policymakers on a continuum of ailments
affecting claimants’ capability to engage in paying work and their pain and
discomfort in doing so (Diamond and Sheshinksi, 1995). Beyond the subset of
clearly incapacitating medical and mental disorders, the extent of “disability” is
ultimately a variable determined by policy.
One approach to evaluating to what extent Disability Insurance leads workcapable individuals to seek benefits in lieu of working is to ask what share of DI
applicants would be working in the absence of the program. Commencing with a
1989 paper by John Bound, a series of studies has attempted to evaluate the work
capability of DI applicants by comparing the labor supply, later in life, of those
accepted and rejected for disability benefits. In these studies, the fraction of
applicants for DI who return to work after their applications is taken as an estimate
of the work capacity of rejected disability applicants, who in turn are presumably
more work-capable than accepted applicants.
This methodology, while ingenious, has important limitations. First, it may be
biased towards underestimating the labor supply disincentives of the Disability
Insurance system (Parsons, 1980, 1991; Bound, 1991). Some rejected applicants
may remain out of the labor force because they are reapplying for DI while other
rejected applicants may be unable to find re-employment because their skills and
opportunities deteriorated during the application process. Second, this approach is
unable to detect the form of cheating identified above in which individuals who are
not disabled and not labor force participants—perhaps because they lack marketable skills—file false claims. These individuals would not be expected to work even
absent a disability award.
Bound’s original study, using data from the 1970s, concluded that at most
30 percent of rejected male applicants for Disability Insurance over the age of 45
would have worked if it were not for the availability of DI benefits. Given the
considerable expansion in DI enrollments since 1984, one might expect the
fraction of rejected applicants who are work-capable would have risen since
Bound’s analysis. Surprisingly, Chen and Van Der Klauuw (2005) find that for
males over age 45 applying for DI benefits during the 1990s, labor supply would
only have been 30 to 40 percent higher were it not for the availability of the
program—a figure that is comparable to both Bound’s earlier estimates and to
estimates by Bound, Burkhauser and Nichols (2001).
The fact that rejected Disability Insurance applicants are not substantially
more likely to return to work in the 1990s than they were in the 1970s suggests one
of two explanations. Either rejected applicants are not more work-capable than in
past years or, alternatively, labor market conditions for those most likely to apply for
disability benefits—primarily low-skilled workers— have deteriorated sufficiently
that even applicants who would have been work-capable several decades ago are
now unlikely to find employment (Chen and van der Klauuw, 2005; Juhn, Murphy
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
87
and Topel, 2002). Consistent with this latter view, Autor and Duggan (2003)
demonstrate that the responsiveness of DI applications to adverse labor market
shocks rose sharply between 1984 and 1998, particularly for less-educated workers.
This pattern suggests that a growing fraction of discouraged and displaced workers
are seeking disability benefits.8
Is the Screening Process Broken?
Some abuse of a large public insurance system like the Disability Insurance
program is inevitable. But has such abuse reached unsustainable levels so that the
DI screening process is effectively broken? In our assessment, the answer is yes. At
an operational level, the Social Security Administration has become progressively
less effective at rejecting claims that fail to meet its selection criteria. At a more
fundamental level, the definition of disability that Congress adopted in 1984 (and
has since expanded) is so encompassing that the DI program appears in practice to
function like a nonemployability insurance program for a subset of beneficiaries,
rather than (primarily) as an insurance program for medical impairment. We
consider these failings in turn.
The 1984 liberalization of Disability Insurance vastly increased the complexity
and subjectivity of disability screening (SSAB, 2001). Prior to 1984 (as discussed
earlier), determinations of disability focused primarily on medical criteria and gave
limited credence to nonverifiable symptoms such as pain and mental disorders.
The revamped process is much more subjective, requiring the Social Security
Administration to evaluate an applicant’s workplace function and the extent of
pain or mental illness, to determine the veracity of supporting evidence provided
by the claimant, and to give primary weight to that evidence unless it is at odds with
other information.
Despite repeated efforts by the Social Security Administration to improve the
efficiency, accuracy and consistency of disability screening (United States General
Accounting Office, 1997; 2003), the disability determination process appears to be
evolving from a bureaucratic function to an adversarial process relying heavily on
appeals and adjudication. In 1986, about half (54 percent) of all applications for
disability who were denied at the initial appeal stage by their state-level Disability
Determination Services Office were subsequently appealed to Administrative Law
Judges. Eleven years later, 83 percent of all reconsideration denials were appealed
in this next step (U.S. GAO, 1997). The Social Security Administration (SSA)
adjudication system is now thought to be the largest system of trial-type adjudication in the world (Verkuil and Lubbers, 2002), employing at present 980 full-time
judges. Of all appeals filed in 2005, 92 percent were for disability cases and only
8 percent of appeals were for retirement or Medicare cases (SSA Annual Statistical
Supplement, 2005).
8
For further discussion of the impact of the Disability Insurance program on labor supply, see Bound
and Waidmann (1992), Bound and Burkhauser (1999), Burkhauser and Daly (2002), Autor and Duggan
(2003) and Stapleton and Burkhauser (2003).
88
Journal of Economic Perspectives
The sophistication of claimant appeals has risen demonstrably. Between 1977
and 2000, the share of disability claimants represented by an attorney at the
Administrative Law Judge hearing nearly doubled from 37 to 70 percent (SSAB,
2001). The share of claimants represented by a medical or vocational expert also
increased sharply, more than quadrupling between 1984 and 2000, from 15 percent
to 65 percent. Claimants represented by an attorney prevail three times as often at
trial as unrepresented claimants (U.S. GAO, 2003). This relationship may not be
entirely causal— claimants with stronger cases may find it easier to obtain representation— but it appears unlikely to be entirely spurious.9 Disability litigation
practice has matured as a viable business model, with hundreds of law firms now
pursuing this specialty. In 1997, the most recent year for which figures were
available, the Social Security Administration paid nearly half a billion dollars to
claimants’ attorneys (SSAB, 1998).
Decisions about what constitutes a legitimate disability claim are gradually
being taken from the Social Security Administration. In 1979, approximately one in
eight Disability Insurance awards (12 percent) was granted by an Administrative
Law Judge. By 2002, this share had more than doubled to 27 percent (House Ways
and Means Committee, 1998, 2004). Nor does the appeal process necessarily stop
at the Administrative Law Judge. In fiscal year 2000, 98 percent of applicants whose
claims were denied by an Administrative Law Judge appealed their claim to the
Social Security Administration Appeals Council, which allowed 2 percent of claims
and remanded another 22 percent back to the judge (60 percent of which will likely
be awarded on remand). Of those denied by the Appeals Council, 13 percent
appealed to the federal civil courts. Between 1990 and 2000, the number of
disability cases filed in federal courts nearly tripled from approximately 5,000 to
15,000, representing 5.9 percent of all civil cases in the Court of Appeals in 2000
(Verkuil and Lubbers, 2002, p. 9). Accounting for all levels of appeal, more than
38 percent of awards to individuals who applied for Disability Insurance between
1997 and 2000 were made after an initial denial, versus just 20 percent between
1975 and 1977 (Lando, Farely and Brown 1982).
Perhaps such a transformation of the screening process was inevitable given
the litigiousness of American society and the large awards at stake. It seems unlikely,
however, that the problem would be as severe were it not for the 1984 reforms.
9
Lawyers in this specialty have strong financial incentives to prolong disability cases until they reach the
appeals phase. Following a successful disability appeal, the Social Security Administration grants 25
percent of the retroactive disability award to the claimant’s attorney (currently capped at $5,300).
Because the size of the retroactive benefit—and hence the attorney’s fee—increases with the duration
of the appeal, it is widely believed that claimant attorneys withhold key evidence (such as medical
records) until the case reaches appeals (U. S. GAO, 1997; Verkuil and Lubbers, 2002). The U. S. General
Accounting Office (1997) reports that in 1997, 76 percent of appealed cases contained new evidence.
The introduction of new evidence late in the appeal is permissible because of an unusual feature of the
Social Security Administration appeals process: the evidentiary record is not closed prior to the appeal
to the federal court.
David H. Autor and Mark G. Duggan
89
Figure 4
Disability Insurance Applications per Nonelderly Adult and the U.S. Unemployment Rate, 1978 –2004
15
8
7
13
6
12
11
5
10
Unemployment rate
DI applications per 1,000 persons aged 25–64
14
4
9
3
8
7
1978
2
1982
1986
1990
1994
1998
2002
Sources: Social Security Administration Office of the Chief Actuary; U.S. Census Bureau; Bureau of
Labor Statistics.
Whatever the reason, the increase in the fraction of DI awards made on appeal
raises the concern that the Social Security Administration is gradually losing
control of the decision-making process on eligibility for disability benefits.
Disability Insurance or Unemployability Insurance?
Disability Insurance applications typically rise and fall with the unemployment
rate, as shown in Figure 4 (Rupp and Scott, 1998; Black, Daniel and Sanders, 2002;
Autor and Duggan, 2003). (The one exception to this pattern, seen during 1980 to
1984, occurred when the retrenchment of DI from 1980 to 1984 reduced disability
applications while the U.S. unemployment rate soared.) Since it seems unlikely that the
onset of disability is countercyclical, this pattern could be taken as prima facie evidence
that the DI system is being abused as a form of unemployment insurance. But from a
legal standpoint, this inference is incorrect. The original, statutory definition of a
disability—“the inability to engage in a substantial gainful activity in the U.S.
economy”— depends explicitly on an applicant’s job opportunities. As Congress has
broadened the scope of the disability program over five decades to encompass workers
of all ages, to cover medium-term (lasting at least one year) as well as permanent
disabilities, and to place substantial weight on both functional and medical criteria, this
focus on labor market criteria has become paramount.
90
Journal of Economic Perspectives
Four of the five steps in Social Security’s current evaluation process ask whether
the Disability Insurance applicant is currently working, can perform basic work tasks,
could return to her past work or could be retrained for other work in the national
economy. By contrast, only one of the five steps concerns whether the applicant’s
impairment meets the Social Security Administration’s (SSA’s) medical criteria. If this
level of medical severity is not met, however, the claim may still be awarded if the SSA
determines that the applicant is unable to perform her past work and is unable to be
retrained for new work (as explained at 具http://www.ssa.gov/dibplan/dqualify5.htm典).
These criteria ensure that adverse economic shocks will spur those with relatively poor
health and limited labor market opportunities to seek and qualify for disability benefits.
To give one illustrative example of the degree to which Disability Insurance has
moved from its original role of insuring workers against physical incapacity to one
of insuring workers against nonemployability, consider what occurred following
passage of a 1996 law disqualifying Disability Insurance claimants whose primary
impairment was drug or alcohol addiction. The Social Security Administration soon
terminated approximately 130,000 beneficiaries (note the spike in Medical Disqualifications in 1997 in Figure 2). Remarkably, two-thirds of these terminated claimants eventually re-qualified for DI benefits under a different impairment (Lewin
Group, 1998). Beyond highlighting the inability of the Social Security Administration to deflect aggressive claimants, this example points to a structural problem.
Because the legislative definition of disability turns primarily on a claimants’
nonemployability, it is ultimately unsurprising that the majority of terminated
beneficiaries would requalify for benefits under a different impairment.
Implications for Future Enrollment and Finances
How Much Further Will Disability Insurance Enrollment Increase?
In the 1980s and 1990s, the rising number of recipients of Disability Insurance
has largely been driven by a steady increase in the award rate per insured population. But in recent years and into the future, the predominant cause of the current
and projected future increase in enrollment is the decline in the exit rate from the
program. As shown earlier in Figure 2, the fraction of DI recipients leaving the
program in each year has consistently fallen. In 1983, 16.4 percent of those
receiving disability benefits in that year exited the program because of death,
entering retirement, or no longer meeting the criteria for disability. By 2005, only
7.2 percent of those receiving disability benefits in that year exited the program.
If one approximates the average duration of time spent receiving disability
benefits as equal to the reciprocal of the exit rate, the mean duration of disability
benefits has increased from six years in 1983 to 14 years in 2004. Assuming that this
duration remains constant and that the DI award rate remains at approximately
0.5 percent per year (the average from 2002 to 2004), in the future the DI program
will ultimately level out when it pays benefits to 7 percent of the nonelderly adult
population. This would represent a 71 percent increase above the 4.1 percent rate
of DI enrollment at the end of 2005.
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
91
For comparison, the Board of Trustees of the Social Security Administration
(2005) makes an “intermediate” projection that Disability Insurance enrollment
will increase from 6.2 million in 2004 to 9.8 million by 2025. However, this
projection assumes a rapid slowdown in the growth of the program. For example,
after increasing by 5.3 percent per year from 2000 to 2004, the trustees assume that
the disability rolls will only increase at 3.5 percent per year from 2004 to 2010 and
at 1.6 percent per year from 2010 to 2020. While the trustees’ projections may prove
accurate, their previous estimates have tended to be too low. The 2001 Trustee’s
report predicted an increase in DI enrollment from 5.0 million in 2000 to
6.1 million in 2005. The actual increase during this five-year period was 36
percent higher.
Implications for Social Security and Medicare Finances
The Social Security program paid out $521 billion in cash benefits to approximately 48.4 million individuals during the 2005 calendar year from two separate
trust funds: the Old Age and Survivors Insurance (OASI) Trust Fund and the
Disability Insurance Trust Fund. Each trust fund has three main sources of revenue:
payroll taxes, interest on the accumulated assets in the trust fund, and income from
the taxation of Social Security benefits. Payroll taxes are by far the most important,
accounting for more than 84 percent of the $702 billion in program revenues
during the 2005 calendar year. Since 1990, workers covered by Social Security have
paid a flat 12.4 percent tax on earnings in each year up to a taxable maximum. This
taxable maximum, indexed to average economy-wide wage growth, was equal to
$90,000 in 2005. While the Social Security tax rate has remained unchanged since
1990, the share going to DI has risen. In 1990 the tax rates for the DI and OASI trust
funds were 1.2 and 11.2 percent, respectively, but they are currently set to 1.8 and
10.6 percent.
Spending on Disability Insurance accounted for 10.1 percent of Social Security
spending in 1990, but rose to 16.6 percent in 2005. Absent major changes to the
Disability Insurance program, the current tax rate of 1.8 percent will not suffice to
finance disability benefits, meaning that a larger share of the Social Security tax will
be unavailable for financing retirement benefits.10
Trends in Disability Insurance enrollment have important implications for
Medicare financing as well. During the 2005 calendar year, DI recipients accounted
for 15.4 percent of all Medicare recipients and 15.1 percent of the $325 billion in
Medicare expenditures (United States Department of Health and Human Services,
2005). The share of Medicare spending accounted for by DI recipients has also
increased.
10
The Social Security trust funds continue to pay benefits to Disability Insurance recipients once they
reach the full retirement age and are shifted from the Disability Insurance to the Old Age and Survivors
Insurance rolls. During the 2002 calendar year, approximately one out of every eight retired worker
awards went to 65-year-olds who had been receiving Disability Insurance. If one accounts for these Old
Age and Survivors Insurance expenditures, the fraction of Social Security payments given to past or
current Disability Insurance recipients is much greater than the 16.6 percent share noted above.
92
Journal of Economic Perspectives
How Should Disability Insurance be Reformed?
The rapid growth of Disability Insurance enrollment during the past two
decades appears poised to continue and even to accelerate as the baby boom
cohorts attain their peak disability years, reaching more than 6 percent of the
nonelderly U.S. adult population in the next decade and accounting for more than
8 percent of the federal budget. The rapid growth of Disability Insurance does not
appear to be explained by a true rise in the incidence of disabling illness, but rather
by policies that increased the subjectivity and permeability of the disability screening process. A broadening societal definition of disability and increased willingness
of individuals to seek government benefits may also have played some role, though
we have no direct evidence on these points.
What might be done? There are three ways to reduce the size of the Disability
Insurance program: shorten the average length of time that claimants receive
benefits by encouraging faster exit; tighten the screening process so that fewer
insured workers may qualify; and reduce the incentives for qualified workers to seek
benefits. We discuss these options in turn.
Raising the exit rate from the Disability Insurance rolls has been tried in
various ways, but has not worked well. As discussed earlier, an aggressive increase in
Continuing Disability Reviews during 1980 through 1983 reduced the DI rolls by
approximately 13 percent in three years. But the ensuing public backlash and a
series of court reversals galvanized Congress’s 1984 liberalization of the disability
screening process and arguably led to the present day crisis; in fact, many of the
beneficiaries terminated from 1980 to 1983 were later reinstated on appeal. Similarly, Congress’ attempt to disqualify DI recipients whose primary impairment was
drug or alcohol addiction did not ultimately remove the majority of these recipients
from the rolls. In another example, Congress in 1999 enacted the Ticket-to-Work
program, which extended Medicare benefits for up to eight years to DI beneficiaries who rejoined the workforce and provided beneficiaries with “tickets” to assign
to their vocational providers, who would receive bonuses if the beneficiary returned
to full-time work. However, only 8,900 of the more than 11.4 million tickets issued
to date have been taken up by claimants and their providers, and fewer than 1,400
of these have led to successful workforce integration.11 These experiences suggest
that once awarded, disability benefits are extremely difficult to rescind.
A more promising avenue for slowing program growth may be to reduce
inflows to the Disability Insurance program. One method to stem inflows is to adopt
more rigorous eligibility criteria that emphasize medical rather than vocational
factors, as was done in the late 1970s and early 1980s. However, any reduction in the
award rate for nonmeritorious claims would surely be accompanied by a rise in the
rejection rate for deserving claimants suffering from difficult-to-verify impairments,
11
Statistics on Ticket to Work (accessed on 3/27/2006) are continually updated at http://www.
socialsecurity.gov/work/Ticket/ticket_info.html#TicketTracker.
David H. Autor and Mark G. Duggan
93
such as pain and mental disorders. Although we view this trade-off as potentially
worthwhile, it is not clear that it would be politically sustainable.
Less controversial would be reforms that simultaneously raised the rejection
rate of nonmeritorious claims while increasing the acceptance rate of deserving
claims. Two policy reforms hold this potential. First, the Social Security Administration might be routinely allowed to commission independent medical and vocational evaluations of Disability Insurance claimants during the initial determination. This change would provide a healthy counterbalance to the current
requirement that Disability Determination Services examiners give highest weight
to source evidence submitted by claimants and their selected medical representatives, which seems to invite a highly skewed presentation of facts.12
A second promising step would be for the Social Security Administration to
consider attorney representation at Administrative Law Judge hearings, as the
independent Social Security Advisory Board (2001) has repeatedly recommended.
At present, claimants are typically represented at appeal by legal and medical
advocates who have a financial stake in the claimant’s success. The Social Security
Administration, by contrast, is entirely dependent on the Administrative Law Judge
to protect the claimant’s and the public’s interests simultaneously (U.S. GAO,
1997). Permitting the Social Security Administration to provide a representative or
attorney to the hearings would ameliorate this almost comically lopsided setting, in
which the Social Security Administration currently loses nearly three-quarters of all
appeals.13
An alternative means to stem Disability Insurance inflows is to lower the
number of workers seeking benefits. Individuals with expensive-to-treat but not
otherwise incapacitating disorders are likely to have difficulty in obtaining health
insurance and hence will find the Medicare component of the DI system particularly attractive. In this way, DI may serve as a health insurer of last resort for
otherwise work-capable individuals. Gruber and Kubik (2005) show that potential
applicants for DI coverage who would lose other insurance coverage while awaiting
eligibility for the disability-related Medicare benefit are substantially less likely to
apply for disability than are potential beneficiaries who have stopgap forms of
health coverage. For this reason, successful efforts to close the fissures in the U.S.
health care system might have the additional benefit of reducing applications for
Disability Insurance.
12
Benitez-Silva, Buchinsky and Rust (2004) propose improving disability screening through better
statistical evaluation of claimant evidence. Golosov and Tsyvinski (2006) make a case for asset-testing of
disability claimants to discourage undeserving applicants, who face an incentive to stockpile assets in
anticipation of submitting a claim (given that they would no longer be working and disability benefits
are lower than potential earnings).
13
The Social Security Administration reported to Congress in 1985 that a pilot study found that
Administrative Law Judge awards made in error could be cut by half if the Social Security Administration
was represented at appeal hearings (Hearing before the Select Committee on Aging, House of Representatives, 99th Congress, 1st session, Mar. 18, 1985). In 1986, however, a federal court ruled that
representation for the Social Security Administration, at least as implemented in the pilot, violated
procedural due process (U. S. GAO, 1997, p. 43).
94
Journal of Economic Perspectives
Finally, one radical reform to reduce the incentive to seek disability benefits is
to scrap the current all-or-nothing cash award in favor of a graduated disability scale
similar to that used by the Disability Compensation program in the Department of
Veterans Affairs. Under this program, benefits awarded are an increasing function
of the claimant’s disability, rated on a scale from zero to 100 percent. Claimants
with mild impairments receive smaller payments and therefore have weaker incentives to seek awards. Moreover, because benefit payments in the Disability Compensation program are not conditioned on staying out of the labor force, the
system does not directly reward labor force nonparticipation.
Relative to the current Disability Insurance system, there are, however, two
clear drawbacks to the Disability Compensation model. First, if the disabled could
both work and collect benefits, the pool of workers seeking and obtaining disability
benefits would surely increase. Second, the workforce nonparticipation requirement of the current DI program likely induces favorable self-screening; only the
neediest applicants—those with low earnings capacity and high disutility of work—
are likely to apply. To date, researchers have largely neglected study of the Veteran’s Disability Compensation program, though careful economic analysis of it is
clearly warranted.
There are no easy reforms to Disability Insurance. The disabled are generally
held in high regard by the public— unlike, for example, the standing of welfare
recipients prior to passage of welfare reform in 1996. General Social Survey data
from 1994 showed that 60 percent of the U.S. respondents believed that “we are
spending too much on welfare” (Weaver, Shapiro and Jacobs, 1995). In the same
year, a Kaiser poll found that 58 percent of Americans believed that the federal
government should be spending “more” or “much more” on poor adults who are
disabled (Center for Study of Policy Attitudes, University of Maryland, 1994). But
with the present value of new DI awards and the accompanying Medicare coverage
now exceeding $200 billion per year, and rising, the cost of postponing reforms to
DI may eventually come to appear even more daunting than the cost of facing them
promptly.
y The authors thank Amy Finkelstein, Jim Hines, Timothy Taylor and Michael Waldman for
suggestions that substantially improved the manuscript, and Alex Chernyakov and Kathleen
Cui for excellent research assistantship. Autor acknowledges financial support from the
National Science Foundation (CAREER SES-0239538) and Duggan from the Alfred P. Sloan
Foundation.
Growth in Social Security Disability Rolls: A Fiscal Crisis Unfolding
95
References
Autor, David H. and Mark G. Duggan. 2003.
“The Rise in the Disability Rolls and the Decline
in Unemployment.” Quarterly Journal of Economics. 118:1, February, 157–206.
Autor, David H. and Mark G. Duggan. 2006.
“The Pure Income Effect of Disability Cash
Transfers: Evidence from the Veteran Administration’s Disability Compensation Program.” Unpublished paper, Massachusetts Institute of
Technology.
Autor, David H., Lawrence F. Katz and Melissa
S. Kearney. 2005. “Trends in U.S. Wage Inequality: Re-Assessing the Revisionists.” NBER Working Paper No. 11627.
Benitez-Silva, Hugo, Moshe Buchinsky and
John Rust. 2004. “How Large are the Classification Errors in the Social Security Disability
Award Process?” NBER Working Paper No.
10219, January.
Bhattacharya, J. and Michael Schoenbaum.
2002. “The Value of the Medicare Benefit for
Social Security Disability Insurance Recipients.”
University of Michigan, Economic Research Initiative on the Uninsured Working Paper No. 10,
July.
Black, Dan, Kermit Daniel and Seth Sanders.
2002. “The Impact of Economic Conditions on
Participation in Disability Programs: Evidence
from the Coal Boom and Bust.”American Economic Review. 92:1, 27–50.
Blau, Francine D., Marianne A. Ferber and
Anne E. Winkler. 2002. The Economics of Women,
Men, and Work. 4th edition. Upper Saddle River,
NJ: Prentice Hall.
Bound, John. 1989. “The Health and Earnings
of Rejected Disability Insurance Applicants.”
American Economic Review. 79:3, pp. 482–503.
Bound, John. 1991. “The Health and Earnings
of Rejected Disability Insurance Applicants: Reply.” American Economic Review. December, 81:5,
pp, 1427–34.
Bound, John and Richard Burkhauser. 1999.
“Economic Analysis of Transfer Programs Targeted on People with Disabilities.” Handbook of
Labor Economics. Vol. 3. Orley Ashenfelter and
David Card, eds. Amsterdam: North-Holland.
Bound, John, Richard Burkhauser and Austin
Nichols. 2001. “Tracking the Household Income
of SSDI and SSI Applicants.” University of Michigan Retirement Research Center Working Paper No. 2001-009.
Bound, John and Timothy Waidmann. 1992.
“Disability Transfers, Self-Reported Health and
the Labor Force Attachment of Older Men: Ev-
idence from the Historical Record.” Quarterly
Journal of Economics. 107:4, pp, 1393–1419.
Burkhauser, Richard and Mary Daly. 2002.
“Policy Watch: U.S. Disability Policy in a Changing Environment.” Journal of Economic Perspectives.
Winter, 16:2, pp. 213–24.
Center for Study of Policy Attitudes, University of Maryland. 1994. Poll of 10/13/1994.
Chen, Susan and Wilbert van der Klauuw.
2005. “The Work Disincentive Effects of the Disability Insurance Program in the 1990s.” Unpublished paper, University of North Carolina,
Chapel Hill, September.
Diamond, Peter and Eytan Sheshinski. 1995.
“Economic Aspects of Optimal Disability Benefits.” Journal of Public Economics. 57:1, pp, 1–23.
Duggan, Mark and Scott Imberman. Forthcoming (2006). “Why Are the Disability Rolls
Skyrocketing?” in Health in Older Ages: The Causes
and Consequences of Declining Disability among the
Elderly. David Cutler and David Wise, eds. University of Chicago Press.
Farber, Henry and Helen Levy. 2000. “Recent
Trends in Employer-Sponsored Health Insurance Coverage: Are Bad Jobs Getting Worse?”
Journal of Health Economics. 19:1, 93–119.
Golosov, Mikhail and Aleh Tysvinski. 2006.
“Designing Optimal Disability Insurance: A Case
for Asset Testing.” Journal of Political Economy.
April, 114:2, pp. 257–79.
Gruber, Jonathan and Jeffrey D. Kubik. 1997.
“Disability Insurance Rejection Rates and the
Labor Supply of Older Workers.” Journal of Public
Economics. April, 64:1, pp. 1–23.
Gruber, Jonathan and Jeffrey D. Kubik. Forthcoming (2005). “Health Insurance Coverage
and the Disability Insurance Application Decision.” Journal of Public Economics.
Hennessey, John and Janice Dykacz. 1993. “A
Comparison of the Recovery Termination Rates
of Disabled-Worker Beneficiaries Entitled in
1972 and 1985.” Social Security Bulletin. Summer,
56:2, pp. 58 – 69.
Juhn, Chinhui, Kevin M. Murphy and Robert
H. Topel. 2002. “Current Employment, Historically Contemplated.” Brookings Papers on Economic
Activity. 1, pp. 79 –116.
Katz, Lawrence F. and David H. Autor. 1999.
“Changes in the Wage Structure and Earnings
Inequality,” in Handbook of Labor Economics. Vol.
3A. Orley Ashenfelter and David Card, eds. Amsterdam: Elsevier-North-Holland.
Lando, M., A. Farley and M. Brown. 1982.
“Recent Trends in the Social Security Disability
96
Journal of Economic Perspectives
Insurance Program.” Social Security Bulletin. 45:8,
3–14.
Lewin Group. 1998. “Policy Evaluation of the
Effect of Legislation Prohibiting the Payment of
Disability Benefits to Individuals Whose Disability is Based on Drug Addiction and Alcoholism:
Interim Report.” Unpublished report by the
Lewin Group, Inc., June 21.
Manton, Kenneth G. and XiLiang Gu. 2001.
“Changes in the Prevalence of Chronic Disability
in the United States Black and Non-Black Population Above Age 65, from 1982 to 1999.” Proceedings of the National Academy of Sciences. May 8,
98:11, pp. 6354 –59.
Organisation for Economic Co-operation and
Development. 2003. “Transforming Disability
into Ability: Policies to Promote Work and Income Security for Disabled People.” Paris:
OECD.
Parsons, Donald O. 1980. “The Decline in
Male Labor Force Participation.” Journal of Political Economy. February, 88:1, pp. 117–34.
Parsons, Donald O. 1991. “The Health and
Earnings of Rejected Disability Insurance Applicants: Comment.” American Economic Review. December, 81:5, pp. 1419 –26.
Rupp, Kalman and Charles Scott. 1988. “Determinants of Duration on the Disability Rolls
and Program Trends,” in Growth in Disability Benefits: Explanations and Policy Implications. Kalman
Rupp and David Stapleton, eds. Kalamazoo, MI:
Upjohn Institute for Employment Research, pp.
139 –76.
Social Security Administration. Various years.
Annual Statistical Supplement.
Social Security Administration. Various years.
DI Annual Statistical Report.
Social Security Advisory Board. 1998. “How
SSA’s Disability Programs Can Be Improved.”
Available at 具http://www.ssab.gov/Publications/
Disability/report6.pdf典.
Social Security Advisory Board. 2001. “Charting the Future of Social Security’s Disability Programs: The Need for Fundamental Change.”
Available at 具http://www.ssab.gov/Publications/
Disability/disabilitywhitepap.pdf典.
Social Security Advisory Board. 2003. “The
Social Security Definition of Disability.” Avail-
able
at
具http://www.ssab.gov/documents/
SocialSecurityDefinitionOfDisability_002.pdf典.
Social Security Administration Board of Trustees. 2005. “The 2005 Annual Report of the
Board of Trustees of the Federal Old-Age and
Survivors Insurance and Disability Insurance
Trust Funds.” Available at 具http://www.ssa.gov/
OACT/TR/TR05/tr05.pdf典.
Stapleton, David C. and Richard V.
Burkhauser. 2003. The Decline in Employment of
People with Disabilities: A Policy Puzzle. Kalamazoo,
MI: Upjohn Institute for Employment Research.
United States Census Bureau. 2005. “Current
Population Reports, P23-209: 65⫹ in the United
States: 2005.” Washington, DC: U.S. Government Printing Office.
United States Department of Health and Human Services. “2005 CMS Statistics.” Available at
具http://www.cms.hhs.gov/MedicareMedicaidStatSupp/downloads/2005_CMS_Statistics.pdf典.
United States General Accounting Office.
1997. “Social Security Disability: SSA Must Hold
Itself Accountable for Continued Improvement
in Decision-Making.” GAO-97-102 Washington,
DC: U.S. Government Printing Office. Available
at 具http://www.gao.gov/典.
United States General Accounting Office.
2003. “SSA Disability Decision Making: Additional Steps Needed to Ensure Accuracy and
Fairness of Decisions at the Hearing Level.”
GAO-04-14. Washington, DC: U.S. Government
Printing Office. Available at 具http://www.
gao.gov/典.
United States House of Representatives, Committee on Ways and Means. Various years. Overview of Entitlement Programs. Some recent versions
are available at: 具http://aspe.hhs.gov/2000gb/典
or at 具http://aspe.hhs.gov/2002)典.
Verkuil, Paul and Jeffrey Lubbers. 2002. “Alternative Approaches to Judicial Review of Social
Security Disability Cases: A Report to the Social
Security Advisory Board.” Available at 具http://
www.ssab.gov/Publications/Disability/VerkuilLubbers.pdf典. (Accessed 10/28/2005).
Weaver, R. Kent, Robert Y. Shapiro, and Lawrence R. Jacobs. 1995. “Poll Trends: Welfare.”
Public Opinion Quarterly. 59:4, p. 606 –27.